Professional Documents
Culture Documents
PH Calculation by Differential Conductivity Measurement in Mixtures of Alkalization Agents Marco Lendi
PH Calculation by Differential Conductivity Measurement in Mixtures of Alkalization Agents Marco Lendi
Extended pH
calculation model
Made in Switzerland
www.swan.ch
Water / Steam Analysis Equipment for Saudi Electric Company Power Plant PP10
manufactured by SWAN Systeme AG.
ABSTRACT
Proper measurement of pH is a key factor in corrosion risk surveillance in water-steam cycles. Since it still seems
difficult to measure the sample pH directly with glass electrodes, pH calculation, using the difference between sample
conductivity before and after a strong acid ion exchanger, is a frequently used alternative. The precision and reliability
of this measuring method is well known and proven in water-steam cycles containing one alkalization agent only.
For applications using mixtures of alkalization agents, the pH calculation model has never been verified. We investi-
gated calculation models to predict the precision and limitations of pH calculation by differential conductivity measure-
ment in morpholine-ammonia and ethanolamine-ammonia mixtures.
INTRODUCTION
Proper measurement of pH is a key factor in corrosion risk An increasing number of power plants are using mixtures
surveillance in water-steam cycles. Many power plant of alkalizing agents like ammonia-ethanolamine or ammo-
chemists still consider the reliable and accurate on-line nia-morpholine. This not only raises the question of
measurement of pH with ion selective glass electrodes in whether or not the existing calculation models can be
condensate or feedwater a difficult task. To guarantee a used, but also of how far the calculated pH deviates from
reliable reading, high maintenance effort is required. reality.
A pH calculation algorithm from differential conductivity In the following paper, the specific conductivity, acid con-
measurement is a widely used alternative method for pH ductivity and pH of high-purity water containing two acid-
measurement in low conductivity water. The VGB guide- base pairs (representing the two alkalizing agents), an acid
line (VGB-S-006-00-2012) describes a calculation model and a neutral contaminant as well as carbon dioxide at
with two conductivity measurements before and after a standard temperature were simulated. The simulated
strong acid ion exchanger. The calculation model is limited parameters were compared with the output of SWAN's
to demineralized feedwater which contains only ammonia, extended calculation model and the VGB model. The devi-
sodium hydroxide or lithium hydroxide as an alkalizing ations and possible solutions are discussed as well.
agent [1].
Equations (2) and (3) are converted to equation (4) with the
dissociation constant KW = [H+] · [OH– ]. The pH after the
cation exchange resin is defined as the negative logarithm
of the [H+]CC concentration.
(
([H+]SC)2 · (lH+ – l B+) + [H +] SC · ( l Cl – + l NH+) (8)
Alkalizing agent CB KW
Ammonia 273
· )[H +] CC –
[ H + ]CC
(
) – kSC + KW · (lOH – + lB+) = 0
Sodium hydroxide 243 The quadratic equation (8) is solved to [H+]CC to calculate
the pH before the cation exchanger from SC and CC. B+
Lithium hydroxide 228
is the equivalent ionic conductivity of the alkalizing agent.
Assumption for the Calculation Model • Weak acid (e.g. carbon dioxide)
In part some assumptions were already mentioned to The system defined by these four variables is a polynomial
explain the calculation models. of 5th degree which is solved to the proton concentration.
So the pH of the defined system is known.
• The sample contains only one alkalizing agent.
• The contamination is sodium chloride. To calculate the total conductivity, one has to know the
concentration of all ions in the solution present at the
• The calculation model works for sample pH within 7.5
known pH. For example, carbon dioxide is a weak acid
to 10.5.
and will dissociate in water to form the bicarbonate ion
In the next section, conditions are discussed which violate HCO3– and the carbonate ion CO32– . With respect to the
the assumptions for the calculation model but represent known pH value from step 2, the equilibrium concentration
realistic situations. The VGB and SWAN calculation mod- of each ion is calculated. Therefore, the specific (or total)
els have been tested with the following disturbances: conductivity of the system is calculated.
• Two alkalizing agents in the sample
In the last step, the system is modified to simulate the
• Carbon dioxide and sodium chloride as contaminants equilibria after a cation exchange resin. The following
changes are made:
The advantage of the expanded model used by SWAN is
the integration of sodium chloride as a contaminant and • The two alkalizing agents remain completely in the
the independent treatment of the specific and cation con- cation exchange resin.
ductivities. Regarding this last point, SC and CC are not
• The neutral salt is converted to its corresponding acid.
coupled by a constant factor (1⁄3 in the VGB guideline). This
A contamination of sodium chloride leads to hydrochlo-
is an important fact for the pH calculation below pH 8.5.
ric acid after the cation exchange resin.
• Weak acids remain in the water.
pH MODELING WITH FOUR VARIABLES With this new system, the pH and all ion equilibrium con-
centrations are calculated. With the resulting ion concen-
In Figure 2, the important steps for the pH modeling are
tration, the conductivity after a strong cation exchange
listed. In the first step, all possible variables are defined.
resin is calculated.
The defined variables are:
• Two alkalizing agents and their concentration ranges The calculated specific and cation conductivities are the
input parameter for the VGB and the SWAN calculation
• Neutral salt (e.g. sodium chloride)
models (Figure 3). With respect to the calculated specific
Figure 2: Figure 3:
Steps for correct pH modeling. The calculated specific and cation conductivities were put into
the VGB and SWAN calculation models.
and cation conductivities, the pH was calculated with the and SWAN calculation models. The calculated pH was
VGB and the SWAN calculation models. Additionally, the compared with the simulated pH shown in Figure 4 to cal-
calculated pH was compared with the calculated pH val- culate the deviation, expressed as ⌬pH.
ues of the differential conductivity calculation systems to
plot the deviations. In Figures 6 and 7, the VGB and SWAN pH calculation
models are compared for one alkalizing agent only and a
contamination of 50 µg · kg–1 sodium chloride. In Figure 6,
the deviation of the VGB model becomes relevant at
COMPARISON OF THE pH CALCULATION ammonia concentrations lower than 1 mg · kg–1. For con-
MODELS ditions with ethanolamine only, the VGB model's result is
0.05 pH units higher than the simulated pH.
1. Mixture of ammonia and ethanolamine with sodium
chloride contamination
Figures 8 and 9 demonstrate that the VGB and SWAN
In the first example, three variables are used: calculation models perform well even in a mixture of
ammonia and ethanolamine and with a simulated sodium
• Alkalizing agent no. 1: Ammonia with a concentration of
chloride contamination of 50 µg · kg–1. The only exclusion
0 mg · kg–1 to 2 mg · kg–1
zones are the limits of the pH calculation models, which
• Alkalizing agent no. 2: Ethanolamine with a concentra- means below a pH of 8 and in regions where one alkalizing
tion of 0 mg · kg–1 to 2 mg · kg–1 agent is present.
• Contamination with 50 µg · kg–1 sodium chloride
In Figures 4 and 5, the simulated pH and specific conduc- 2. Mixture of ammonia and morpholine with sodium
tivity before the cation exchange resin are shown with chloride contamination
respect to the ammonia and ethanolamine (ETA) concen-
This simulation follows the first example, but with morpho-
trations. The two figures summarize the results of steps 2
line as the second alkalizing agent. Additionally, the con-
and 3 of the modeling process shown in Figure 2.
centration ranges of the alkalizing agents have been
increased to 4 mg · kg–1 to get closer to real dosing condi-
Assuming that both alkalizing agents remain in the cation
tions in French nuclear plants [5].
exchange resin and sodium chloride is exchanged for
hydrochloric acid, the strong acidic solution has a • Alkalizing agent no. 1: Ammonia with a concentration of
simulated pH of 6.062 and a cation conductivity of 0 mg · kg–1 to 4 mg · kg–1
0.369 µS · cm–1.
• Alkalizing agent no. 2: Morpholine with a concentration
of 0 mg · kg–1 to 4 mg · kg–1
The simulated specific and cation conductivities were
used to re-calculate the pH of the solution with the VGB • Contamination with 50 µg · kg–1 sodium chloride
10
10.0
9
k [µS · cm –1]
pH
5.0
8
7 0
2.0 2.0 2.0 2.0
1.5 1.5 1.5 1.5
ETA 1.0 1.0 –1 ] 1.0 1.0 –1 ]
kg
mg ·
[mg 0.5 ETA kg
· kg –1 0.5 [ [mg 0.5 0.5 [ mg ·
] 0 0 NH 3 · kg –1 0 0 NH 3
]
Figure 4: Figure 5:
pH before cation exchange resin. Specific conductivity in mixture of NH3 and ETA.
0.02 0.02
DpH
DpH
0 0
– 0.02 – 0.02
– 0.04 – 0.04
Figure 6: Figure 7:
Deviation with one alkalizing agent only (NH3). Deviation with one alkalizing agent only (ETA).
0.05 0.05
DpH
DpH
0 0
–0.05 –0.05
2.0 2.0
1.5 1.5
] –1 ]
–1 1.0
1.0
k g · kg
0 0.5 g· 0
0.5
0.5
[mg
0.5 1.0 [m 1.0 0 A
1.5 2.0
0
ET
A
NH3 [mg · kg –1
1.5 2.0 ET
NH3 [mg · kg –1 ]
]
Figure 8: Figure 9:
pH deviation of VGB calculation model. pH deviation of SWAN calculation model.
The pH and specific conductivity with respect to the alka- line instead of ammonia and vice versa – is run, SWAN's
lizing agent mixture are simulated in Figures 10 and 11. calculation gives an error not higher than 0.04 pH units.
After a cation exchange resin, the pH and cation conduc-
tivity are the same as in the first example (pH = 6.062,
= 0.369 µS · cm–1). 3. Mixture of ammonia and ethanolamine with carbon
dioxide contamination
In Figures 12 and 13, the VGB and SWAN pH calculation
In this model, the contamination source is a weak acid
models are compared with the simulated pH value of the
(carbon dioxide reacts with water to form carbonic acid).
system. The VGB model does not include morpholine, this
The concentration ranges of the alkalizing agents are
explains why the error increases at low ammonia concen-
comparable with those from example no. 1.
trations. The opposite behavior is shown with the SWAN
model, which gives the best results with low ammonia con- • Alkalizing agent no. 1: Ammonia with a concentration of
centrations. Even if the wrong alkalizing agent – morpho- 0 mg · kg–1 to 2 mg · kg–1
10 15.0
9 10.0
k [µS · cm –1]
pH
8 5.0
7 0
4.0 4.0 4.0 4.0
3.0 3.0 3.0 3.0
Mor 2.0 Mor 2.0 2.0 –1 ]
2.0 pho
g ]
–1
pho
line 1.0 · k line 1.0 1.0
[mg
· kg
[mg 1.0 [mg [mg
· kg –1 0 0 NH 3
· kg – 1 0 0 NH 3 ]
]
0.02 0.02
DpH
DpH
0 0
– 0.02 – 0.02
– 0.04 – 0.04
• Alkalizing agent no. 2: Ethanolamine with a concentra- increases the cation conductivity from 0.055 to
tion of 0 mg · kg–1 to 2 mg · kg–1 0.214 µS · cm–1 after the cation exchange resin.
• Carbon dioxide (CO2) contamination of 50 µg · kg–1
In Figure 15, the influence of CO2 on the pH calculation of
Sources of carbon dioxide intake are air leakages and a system containing only ammonia is shown. The devia-
decomposition products of organic substances in the tion is a consequence of the assumption that an increased
water-steam cycle. Carbon dioxide is highly soluble in cation conductivity is treated as sodium chloride contami-
alkaline water because it reacts with water to form car- nation. This error becomes more important with lower
bonic acid ions. In Figure 14, it is shown that the pH of the ammonia concentrations. Although the properties of CO2
water decreases because of the weak acid. The carbonic are completely different from sodium chloride, the influ-
acid ions will pass the cation exchange resin, giving a low ence on pH calculation based on differential conductivity
pH of 6.271 and an increased cation conductivity of measurement is negligible.
0.214 µS · cm–1. In other words, the 50 µg · kg–1 CO2
10
– 0.01 VGB model
SWAN model
9 – 0.02
pH
DpH
8 – 0.03
7 – 0.04
2.0 2.0
1.5 1.5
ETA 1.0 1.0 –1 ]
[mg 0.5 · g
k
· kg –1 0.5 0 H 3
[mg – 0.05
0 0.5 1.0 1.5 2.0
] 0 N
NH3 [mg · kg –1 ]
Degassed cation conductivity (the sample after a strong regions where the deviation is between 0.00 and 0.02 pH
acid ion exchanger is degassed to remove the carbon units, yellow from +0.02 to 0.03, orange from 0.03 to 0.04
dioxide) is a powerful tool to eliminate the influence of CO2 and red is for deviations +0.04 and higher. The scale for
on the cation conductivity. In this example, 50 µg · kg–1 negative deviations uses bluish colors accordingly.
CO2 increased the cation conductivity from 0.055 to
0.214 µS · cm–1. However the influence of the too high It is clear that the VGB calculation model gives slightly
cation conductivity reading on the pH calculation is hardly better values with high ethanolamine concentrations.
measurable. However with lower alkalizing agent concentrations, the
pH values are too low. Therefore, the SWAN calculation
returns better results with lower alkalizing agent concen-
4. Mixture of ammonia and ethanolamine with carbon trations or if the sample pH is below 8. In summary, the
dioxide and sodium chloride contamination deviations from the VGB and SWAN models are lower
than 0.05 pH units. In consideration of the conditions in
In this example, ammonia and ethanolamine are combined
the water-steam cycle (pH between 8 and 10.5), the devi-
with two contaminations.
ations are lower than 0.02 pH units.
• Alkalizing agent no. 1: Ammonia with a concentration of
0 mg · kg–1 to 2 mg · kg–1
• Alkalizing agent no. 2: Ethanolamine with a concentra-
tion of 0 mg · kg–1 to 2 mg · kg–1
• Carbon dioxide (CO2) contamination of 50 µg · kg–1 CONCLUSION
• Contamination with sodium chloride of 50 µg · kg–1 Two pH calculation models based on differential conduc-
–1 tivity measurement were compared with a simulated pH in
With a contamination of 50 µg · kg sodium chloride and
mixtures of alkalizing agents. Additionally, contaminations
50 µg · kg–1 CO2, the simulated pH after the strong acid
with sodium chloride and carbon dioxide were simulated
ion-exchange resin is 5.929 with a cation conductivity of
to model the conditions in a water-steam cycle as realisti-
0.491 µS · cm–1.
cally as possible.
The simulated model represents conditions which are
On the basis of realistic simulated water-steam cycle con-
close to or higher than the limitations of cation conductiv-
ditions, the following conclusions can be drawn for mix-
ity in water-steam cycles. Figures 16 and 17 show the pH
tures of alkalizing agents:
deviation of the much simpler calculation model based on
differential conductivity compared with that of the com- • In a pH range from 8 to 10.5, the deviation of both pH
plex pH simulation model. The green color indicates calculation models was lower than 0.02 pH units.
0.04 0.04
0.02 0.02
DpH
DpH
0 0
–0.02 –0.02
2.0 2.0
–0.04 –0.04
1.5 1.5
–1 ] –1 ]
1.0 1.0
k g · kg
0 0.5 g· 0 0.5
0.5 mg
0.5 1.0 [m 1.0 0 A
[
0 A
NH3 [mg · kg –1
1.5 2.0 ET NH3 [mg · kg –1
1.5 2.0 ET
] ]
• In samples with a pH lower than 8.5 and a high con- THE AUTHORS
tamination source, SWAN's expanded pH calculation
Marco Lendi (B.S., Chemistry, University of Applied
model gives better results than the VGB calculation.
Science, Zurich, Switzerland) joined SWAN Analytical
• Deviation in pH calculation due to carbon dioxide as a Instruments in 2009 working first as a quality manager and
contaminant is insignificant. starting in 2010 in SWAN's customer support group. In
2012 he was promoted to the R&D group.
pH calculation based on differential conductivity measure-
ment is a reliable and low maintenance alternative to ion Heinz Wagner (Ph.D., Physical Chemistry, University of
selective pH measurement even in samples containing Zurich, Switzerland) has a 35-year background in research
mixtures of alkalizing agents. and development for optical systems for industrial
process applications in gas and water analysis. He held
positions as the head of R&D for photometric gas analysis
for Tecan AG (Switzerland) and in online spectrometry with
REFERENCES Optan AG (Switzerland) before joining SWAN Analytical
Instruments in 2003 as a research scientist for water qual-
[1] Sampling and Physico-Chemical Monitoring of Water
ity control applications.
and Steam Cycles, 2012, VGB PowerTech Service
GmbH, Essen, Germany, VGB-006-00-2012-09-EN Peter Wuhrmann (Ph.D., Analytical Chemistry, Swiss
Federal Institute of Technology ETHZ, Zurich, Switzerland)
[2] Maurer, H., On-Line pH-Measurement by Differential
conducted several years of research work with ion-sensi-
Cation and Specific Conductivity, 1997. Presented at
tive microelectrodes for intracellular ion measurements at
the 1997 International Chemistry On-Line Process
the Institute of Cell Biology at the Swiss Federal Institute
Instrumentation Seminar in Clearwater Beach, FL,
of Technology (ETHZ). He was a member of the founding
U.S.A.
group of SWAN Analytical Instruments, where he has been
[3] Bursik, A., PowerPlant Chemistry, 2009, 11(4), 220. responsible for R&D since 1991.
[4] Wagner, H., PowerPlant Chemistry, 2012, 14(7), 455.
CONTACT
[5] Nordmann, F.; Aspects on Chemistry in French
Nuclear Power Plants, 2004. Presented at the 14th Marco Lendi
International Conference on the Properties of Water SWAN Analytical Instruments AG
and Steam in Kyoto, Japan. P.O. Box 398
8340 Hinwil
Switzerland
E-mail: marco.lendi@swan.ch
Headquarters:
SWAN Analytische Instrumente AG
8340 Hinwil/Switzerland
Phone +41 44 943 63 00
swan@swan.ch
Worldwide representation...
South Africa
Germany
France
Great Britain
Italy
Austria
Turkey
Spain
China
Hong Kong
Singapore
Korea
USA
United Arab Emirates